You are on page 1of 14

AUTONOMOUS QUADCOPTER

SUMMER PROJECT '13


PROJECT MENTORS:
Nikhil Upadhye
Ayush Mittal
Sonu Agarwal

TEAM MEMBERS:
Kartik Agarwal
Karthik Korada
Piyush Sahoo
Rushikesh Chaudhari
Vikram Singh

INTRODUCTION
A Quad copter is a multicopter lifted and propelled by four

rotors. In this project we aimed to build an Autonomous Quad


copter which can balance itself while flying. This quad copter
also consists of manual control system (transmitter-receiver)
.For our project we decided to build the quad copter frame in a
simpler way having symmetrical four arms on which a motor
with a propeller is mounted on every arm.
To make this Quad copter autonomous, we used Arduino
platform to program and applied PID algorithm to calculate the
output values of motor commands by using input values from
transmitter and sensors. We used an Inertial Measurement Unit
(IMU) sensor which give values regarding angles and angular
velocities of quad copter frame.

MECHANISM
Each rotor produces a thrust and a torque about its center of
rotation and these forces are used to fly and move Quad
copter. Two rotors mounted on opposite arms of quad copter
are set into clockwise and another two anticlockwise. These
orientation of motors and their direction of rotation cancels all
the torque generated given the speed of the motors are same.
Now, if we change the speed of the motors attached on the left
and right of quad keeping the speed of the other two same,
creates 'Yaw' motion. Similarly, 'Pitch' and 'Roll' movements are
gained by changing the speed of different motors. See the
images below

WHAT IS AUTONOMOUS?
In general, Quad copter are very much difficult to control by
only manual control without using any sensors. So, if we use
some specific motion sensors and apply some algorithms to the
values read from these sensors, the Quad copter can itself
balance its motion. If we use the above technique to control
the Quad copter, it can then be called as "Autonomous".

MANUAL COTROL
All the motors are connected to Electronic speed controllers
(ESC) which control the speed of the rotors and supply motor
the required power. In the manual control, the signals to the

ESC's are directly sent by the transmitter remote control


through receiver. Transmitter is used by the user to control the
quad manually. If the user changes the input, transmitter sends
radio signals to receiver and receiver changes these signals to
PWM signals. These PWM signals are sent to ESCs causing the
change in the speed of the motors. For our project, we are
using 'Futaba' transmitter and receiver.

OUR APPROACH
In our project, we decided to use Arduino Mega 2560 as our
programmer and IMU sensors (9 DOF and 6 DOF IMU) as the
motion sensors. We decided to use PID algorithm in our code to
get the final motor commands. There are different modes of
PID algorithms which can be used to control a quad copter such
as Stabilize, Attitude and GPS control. We first aimed to use
gyro rates in our PID algorithm i.e. the Quad copter can itself
reduce vibrations generated in its frame due to non-zero

angular velocities

to acquire a stable current position.

IMU
We have used both "9 DOF razor" and "6 DOF" IMU sensors to
get values of angles changed from initial position along all the
three axis and angular velocities of the same. These values are
used as input values to our arduino code. Generally an IMU
consists of an accelerometer, gyroscope and magnetometer (in
case of 9 DOF).The values read from this components are
calibrated to get the desired sensor values. The libraries used
for reading calibrated data from 9 DOF and 6 DOF are "RAZOR
AHRS" and "FreeSixIMU" respectively.

CODING and PID algorithm


We have done all our programming on arduino mega in C / C++
language. Arduino code consists of a setup part and a loop
which runs continuously containing the calculations part. The
main part of program of Quad copter was PID algorithm. PID is
used for minimizing the error between the target position and
the current position the quad has achieved. PID algorithm is
basically Proportional-Derivative Integral controller used for
loop feedback mechanism. PID consists of mainly three
constants 'P', 'I' and 'D'. The 'P' term produces an output value
that is proportional to the current error value. The 'I' term is
proportional to both the magnitude of the error and duration
of error. The last 'D' term is used to calculate the derivative of
the process error. We wrote the code to read the signals from
the RC receiver using "PinChangeInt" library. We also wrote the
code to read sensor values. Please see the "IMU" section. Then

we wrote a function that includes receiver, sensor and


calculation of PID algorithm. Then we formed a loop that runs
approximately on 100 Hz and continuously does the
calculations. We also wrote a function that writes motor
commands to the individual motors in terms of PWM signals.

INTEGRATING ALL THE FACTORS


At first we had a wooden frame on which four motors with

propellers were mounted. The arduino circuit board along with


IMU sensors and RC receiver was mounted on the centre of
quad frame. Then we uploaded the code and tested each
component on the quad frame to check if its working properly.
At first we checked the basic working of PID algorithms through
'Roll, Pitch' and 'Yaw' motion. Then we tested for PID tuning
setting one by one the 'P', 'I' and 'D' terms by checking its
different motion behaviour. It was a step by step process to get
to the final testing phase.

PROBLEMS FACED
1) The data reading frequency of 9 DOF was too much lower to
run the code perfectly.9 DOF communicates with Arduino
through SPI communication which is relatively slower than I2C
communication used in 6 DOF IMU. Because of this we decided
to use 6 DOF IMU which gives data at approximately 333 Hz.
2) Due memory overload in our main code there was a problem
occured in which arduino code was setting up again and again.
3) We had a problem of breaking of the frame many times
because of our wooden frame. But it reduced the vibrations
effectively allowing the IMU sensors give accurate values.

4) There was a problem in reading signals from the RC


transmitter, we had to find a suitable library function for this
task. Finally we used "PinChangeInt" library.
5) Firstly it was easy to read only one channel of RC Receiver.
But we had a problem in reading multiple channels. To solve
this problem we used values of multiple channels of RC receiver
directly in the PID algorithm.
6) We first used "servo.write" function to give PWM signals to
ESC's which gave lesser accuracy, because it worked only in the

rage of 40-170 degrees. To resolve this problem, we used


"servo.writemicroseconds" function.

RESULTS
Having integrated all the things nicely we have a good result of
our project. We are now able to fly and stabilize of our Quad
copter for 50-60 seconds which is considered as a good flight.
We are using only gyro-rates in our PID codes. We are yet to
use Euler angles used for the horizontal stabilization of the
quad copter by using stabilize PID mode.

FUTURE PLANS
In the future we are looking forward to make our Quad copter
fully autonomous without using any other sensor. Also we are
thinking of using GPS for long range autonomous flight or
ultrasonic sensors for object avoidance. By using these
techniques we can reach to the level of autonomous Quad
copter.

REFERENCES
https://dev.qu.tu-berlin.de/projects/sf-razor-9dof-ahrs/wiki/Tutorial
http://brettbeauregard.com/blog/2011/04/improving-the-beginners-pi
d-introduction/

http://www.starlino.com/imu_guide.html
https://github.com/AeroQuad/AeroQuad/blob/master/Libraries/AQ_M
otors/Examples/MotorTest/MotorTest.ino
http://blog.mobileapes.com/2010/04/control-all-four-motors-from-ard
uino.html
http://rcarduino.blogspot.co.uk/2012/04/how-to-read-multiple-rc-chan
nels-draft.html
http://www.astro.hr/library/Magnetometer.pdf